Method, device and equipment for dynamic speed limit optimization of mixed traffic road and medium

Abstract

The application provides a kind of human-vehicle mixed road dynamic speed limit optimization method, device, equipment and medium, it is related to intelligent transportation technical field, comprising: obtaining the road visual data that vision sensor is collected to human-vehicle mixed road;From the current road feature of road visual data;Based on current road feature and the current environmental data corresponding to human-vehicle mixed road, judge whether to meet dynamic speed limit condition;If yes, then determine initial dynamic speed limit data through dynamic speed limit model;Determine weather correction average speed based on vehicle state data and current environmental data, utilize weather correction average speed to correct initial dynamic speed limit data, obtain target dynamic speed limit data.The application can solve the problem of low traffic efficiency, significant safety hazards and lag of traditional speed limit mode in mixed road scene.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation technology, and in particular to a method, device, equipment and medium for dynamic speed limit optimization on roads with mixed pedestrian and vehicle traffic. Background Technology

[0002] In traffic management on roads where pedestrians and vehicles share the same space, existing technologies mainly rely on fixed speed limits and zoned speed limits. The core features and implementation details of these technical solutions are as follows: 1) Fixed Speed ​​Limits: Fixed speed limit schemes set speed limits during the road design phase based on factors such as road type, pedestrian traffic, and vehicle traffic. For example, in mixed-traffic scenarios such as around schools and commercial districts, the speed limit is usually set at 30 km / h or lower to ensure pedestrian safety. Once the speed limit is set, it is generally not adjusted according to dynamic traffic changes.

[0003] 2) Area speed limits: Area speed limit strategies implement speed restrictions within specific areas and are usually used in conjunction with facilities such as traffic signs, speed bumps, and traffic lights. For example, during peak hours or under special weather conditions, speed limits can be adjusted through manual intervention or preset rules, but this adjustment method lacks real-time and precision.

[0004] 3) Dynamic speed limits based on vehicle speed: Some intelligent transportation technologies propose dynamic speed limit schemes based on vehicle speed monitoring, such as dynamically adjusting speed limits on highways or urban arterial roads by detecting traffic density and average vehicle speed. These schemes usually rely on a single data source (such as geomagnetic sensors or cameras) and cannot comprehensively consider the complex scenario requirements of roads where pedestrians and vehicles share the road.

[0005] Although existing technologies have improved traffic management on mixed-traffic roads to some extent, the following major problems still exist: 1) Lack of real-time and dynamic adaptability: Existing fixed speed limits and zoned speed limits cannot adapt to the dynamic changes in pedestrian traffic, non-motorized vehicle activity range, and motorized traffic flow in roads where pedestrians and vehicles share the road. Especially during peak hours, fixed speed limits may lead to a waste of road resources or exacerbate traffic congestion.

[0006] 2) Limited data and low decision-making accuracy: Current dynamic speed limit technologies mostly rely on a single data source (such as vehicle speed or traffic flow), which does not fully consider the random behavior of pedestrians and non-motorized vehicles in mixed traffic scenarios, making it difficult for speed limit decisions to fully reflect the actual traffic conditions.

[0007] 3) It is difficult to balance safety and efficiency: While improving traffic efficiency, existing solutions often neglect the safety needs of pedestrians and non-motorized vehicles; conversely, overemphasizing safety may lead to a significant reduction in the traffic efficiency of motor vehicles, making it difficult to achieve a balance between the two.

[0008] 4) Lack of feedback optimization mechanism: Existing solutions usually lack the ability to evaluate and optimize the speed limit effect in real time. The speed limit strategy is difficult to dynamically adjust with changes in traffic conditions, resulting in decision lag. Summary of the Invention

[0009] In view of this, the purpose of the present invention is to provide a method, device, equipment and medium for dynamic speed limit optimization on roads with mixed traffic of pedestrians and vehicles, which can solve the problems of low traffic efficiency, significant safety hazards and the lag of traditional speed limit methods in mixed traffic road scenarios.

[0010] In a first aspect, the present invention provides a method for dynamic speed limit optimization on roads with mixed pedestrian and vehicle traffic, wherein multiple visual sensors are deployed in a designated area of ​​the road with mixed pedestrian and vehicle traffic, and the method includes: Acquire road visual data collected by visual sensors for roads with mixed pedestrian and vehicle traffic; The current road features are extracted from the road visual data. The current road features include one or more of the following: vehicle status data, pedestrian status data, road occupancy rate, and probability of pedestrian-vehicle conflict. The probability of pedestrian-vehicle conflict is obtained by processing the road visual data through a graph neural network. Based on the current road characteristics and the current environmental data corresponding to the road with mixed pedestrian and vehicle traffic, determine whether the dynamic speed limit conditions are met. If so, the initial dynamic speed limit data is determined through the dynamic speed limit model; The weather-corrected average speed is determined based on vehicle status data and current environmental data. The initial dynamic speed limit data is then corrected using the weather-corrected average speed to obtain the target dynamic speed limit data.

[0011] In one implementation, extracting current road features from road visual data includes: Based on road visual data, identify the vehicle and pedestrian zones on roads with mixed pedestrian and vehicle traffic, and extract one or more of the following current road features: Target detection is performed on the vehicle area and the pedestrian area respectively to obtain vehicle bounding boxes and pedestrian bounding boxes, and vehicle state data and pedestrian state data are determined based on the vehicle bounding boxes and pedestrian bounding boxes; Road occupancy is determined based on the boundary information of vehicle traffic areas and pedestrian areas; The probability of human-vehicle conflict is determined based on vehicle bounding boxes and pedestrian bounding boxes using a graph neural network.

[0012] In one implementation, a graph neural network is used to determine the probability of vehicle-pedestrian conflict based on vehicle bounding boxes and pedestrian bounding boxes, including: Using vehicle bounding boxes as vehicle nodes and pedestrian bounding boxes as pedestrian nodes, and assigning node features to vehicle nodes and pedestrian nodes, an initial mixed-traffic graph structure is obtained. The initial node features include one or more of the following: speed features, position features, movement direction features, and node type features. If the distance between any two nodes is less than a preset threshold, an edge is established between the two nodes to obtain the intermediate pedestrian and vehicle mixed traffic graph structure. Perform at least one message pass on the intermediate human-vehicle mixed traffic graph structure to determine the node aggregation characteristics of vehicle nodes and pedestrian nodes, and obtain the target human-vehicle mixed traffic graph structure. Conflict event identification results are generated based on the target pedestrian-vehicle mixed traffic graph structure using a graph neural network. The number of conflict events in the conflict event identification results is counted, and the ratio between the number of conflict events and the total number of conflict event identification results is used as the probability of human-vehicle conflict.

[0013] In one implementation, based on current road characteristics and current environmental data corresponding to roads with mixed pedestrian and vehicle traffic, it is determined whether dynamic speed limit conditions are met, including: If the current road characteristics and the current environmental data corresponding to the mixed pedestrian and vehicle road meet any one of the following sub-conditions, the dynamic speed limit condition is determined to be met: Sub-condition 1: The probability of a human-vehicle conflict is greater than the first preset probability threshold; Sub-condition 2: The speed difference between the maximum and minimum vehicle speed data in the vehicle status data is greater than a preset speed difference threshold; Sub-condition 3: The pedestrian density in the pedestrian status data is greater than the first preset density threshold; Sub-condition 4: The snowfall or rainfall intensity data in the current environmental data is greater than the preset intensity threshold; Sub-condition 5: The visibility data in the current environment is less than the preset visibility threshold.

[0014] In one implementation, the dynamic speed limit model includes a safe speed limit sub-model, a traffic saturation speed limit sub-model, and a pedestrian priority speed limit sub-model; the initial dynamic speed limit data is determined through the dynamic speed limit model, including: Based on the current environmental data, determine the weather correction factor and the visibility correction factor respectively; The safe speed limit, the flow saturation speed limit, and the pedestrian priority speed limit are determined by the safe speed limit sub-model, the flow saturation speed limit sub-model, and the pedestrian priority speed limit sub-model, respectively. By using weather correction factors and visibility correction factors, the minimum values ​​among the safe speed limit, traffic saturation speed limit, and pedestrian priority speed limit are corrected to obtain the initial dynamic speed limit data.

[0015] In one implementation, determining the weather-corrected average vehicle speed based on vehicle status data and current environmental data includes: The weather-corrected average speed is calculated by multiplying the average speed data in the vehicle status data, the weather correction factor, and the visibility correction factor.

[0016] In one implementation, the initial dynamic speed limit data is corrected using weather-corrected average vehicle speed to obtain target dynamic speed limit data, including: Determine the current vehicle speed threshold based on initial dynamic speed limit data; If the weather-corrected average vehicle speed is less than the current vehicle speed threshold, the probability of pedestrian-vehicle conflict is less than the second preset probability threshold, and the pedestrian density in the pedestrian status data is less than the second preset density threshold, the initial dynamic speed limit data will be adjusted according to the preset increment until the weather-corrected average vehicle speed reaches the preset road free-flow speed. If the probability of a pedestrian-vehicle conflict is greater than the third preset probability threshold, and the pedestrian density in the pedestrian status data is greater than the third preset density threshold, the initial dynamic speed limit data will be adjusted according to the preset reduction until the adjusted speed limit data reaches the preset minimum safe speed limit.

[0017] Secondly, the present invention also provides a dynamic speed limit optimization device for roads with mixed pedestrian and vehicle traffic. Multiple visual sensors are deployed in designated areas of the mixed pedestrian and vehicle traffic road. The device includes: The data acquisition module is used to acquire road visual data collected by the visual sensor for roads with mixed pedestrian and vehicle traffic. The feature extraction module is used to extract current road features from road visual data. Current road features include one or more of vehicle status data, pedestrian status data, road occupancy rate, and vehicle-pedestrian conflict probability. The vehicle-pedestrian conflict probability is obtained by processing the road visual data through a graph neural network. The judgment module is used to determine whether the dynamic speed limit conditions are met based on the current road characteristics and the current environmental data corresponding to the road where pedestrians and vehicles share the road. The speed limit determination module is used to determine the initial dynamic speed limit data through the dynamic speed limit model when the result of the judgment module is yes. The speed limit correction module is used to determine the weather-corrected average vehicle speed based on vehicle status data and current environmental data, and to correct the initial dynamic speed limit data using the weather-corrected average vehicle speed to obtain the target dynamic speed limit data.

[0018] Thirdly, the present invention also provides an electronic device including a processor and a memory, the memory storing computer-executable instructions executable by the processor, the processor executing the computer-executable instructions to implement any of the methods provided in the first aspect.

[0019] Fourthly, the present invention also provides a computer-readable storage medium storing computer-executable instructions, which, when invoked and executed by a processor, cause the processor to implement any of the methods provided in the first aspect.

[0020] This invention provides a method, apparatus, device, and medium for optimizing dynamic speed limits on mixed-traffic roads. Multiple visual sensors are deployed in a designated area of ​​the mixed-traffic road. First, road visual data collected by the visual sensors for the mixed-traffic road is acquired. Then, current road features are extracted from the road visual data. These features include one or more of vehicle status data, pedestrian status data, road occupancy rate, and the probability of pedestrian-vehicle conflict. The probability of pedestrian-vehicle conflict is obtained by processing the road visual data using a graph neural network. Next, based on the current road features and the corresponding current environmental data for the mixed-traffic road, it is determined whether the dynamic speed limit conditions are met. If so, initial dynamic speed limit data is determined using a dynamic speed limit model. Finally, based on the vehicle status data and the current environmental data, a weather-corrected average vehicle speed is determined. This weather-corrected average vehicle speed is then used to correct the initial dynamic speed limit data to obtain the target dynamic speed limit data. The above method determines whether a speed limit adjustment is triggered based on current road features and current environmental data extracted from road visual data. If so, it determines the initial dynamic speed limit through a dynamic speed limit model, and then uses weather-corrected average vehicle speed to correct the initial dynamic speed limit data to obtain the final target dynamic speed limit data. This invention can alleviate congestion on roads with mixed pedestrian and vehicle traffic by adjusting the speed limit in real time, improve road communication capabilities, and ensure the traffic safety of pedestrians, non-motorized vehicles, and motorized vehicles, thus achieving refined and intelligent management of speed limits on roads with mixed pedestrian and vehicle traffic.

[0021] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention are realized and obtained through the structures particularly pointed out in the description and the drawings.

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0023] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0024] Figure 1 A flowchart illustrating a method for optimizing dynamic speed limits on roads with mixed pedestrian and vehicle traffic, provided in an embodiment of the present invention; Figure 2 A flowchart illustrating another method for optimizing dynamic speed limits on roads with mixed pedestrian and vehicle traffic, provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of a dynamic speed limit optimization device for roads with mixed pedestrian and vehicle traffic, provided in an embodiment of the present invention. Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] Currently, existing technologies suffer from the following problems: lack of real-time and dynamic capabilities; limited data, resulting in low decision-making accuracy; difficulty in balancing safety and efficiency; and lack of feedback optimization mechanisms. Based on this, this invention provides a method, device, equipment, and medium for dynamic speed limit optimization on mixed-traffic roads, which can solve the problems of low traffic efficiency, significant safety hazards, and the lag of traditional speed limit methods in mixed-traffic road scenarios.

[0027] To facilitate understanding of this embodiment, a detailed description of the dynamic speed limit optimization method for mixed pedestrian and vehicle roads disclosed in this invention will be provided first. Multiple visual sensors are deployed in designated areas of the mixed pedestrian and vehicle roads. (See [link to relevant documentation]). Figure 1 The diagram shows a flowchart of a dynamic speed limit optimization method for roads with mixed pedestrian and vehicle traffic. The method mainly includes the following steps S102 to S110: Step S102: Obtain road visual data collected by the visual sensor for roads with mixed pedestrian and vehicle traffic.

[0028] Among them, the visual sensors can be integrated radar-visual cameras. For example, in the straight sections of roads where pedestrians and vehicles share the same road, one integrated radar-visual camera should be installed for every 100 kilometers to ensure that the detection range of each lane and adjacent area is covered. At intersections, at least two high-definition cameras should be installed in each direction to ensure all-round monitoring of the intersection. For complex intersections, the number of devices should be increased by 50% according to the monitoring coverage of the intersection.

[0029] The road vision data includes at least LiDAR data collected by the integrated radar-visual machine and visual capture data collected by high-definition cameras.

[0030] Step S104: Extract current road features from road visual data.

[0031] The current road features include one or more of the following: vehicle status data, pedestrian status data, road occupancy rate, and pedestrian-vehicle conflict probability. Vehicle status data may include traffic flow and speed, pedestrian status data may include pedestrian density, road occupancy rate is the ratio of the area occupied by all vehicles and pedestrians to the total road area, and pedestrian-vehicle conflict probability indicates the probability of a conflict between pedestrians and vehicles. This probability is obtained by processing road visual data using a graph neural network. In one implementation, the road visual data first identifies vehicular and pedestrian traffic areas, and then further detects the vehicle and pedestrian bounding boxes within them to determine the vehicle and pedestrian status data respectively. Simultaneously, the road occupancy rate is determined based on the boundary information of the vehicular and pedestrian traffic areas. A target mixed-traffic graph structure is constructed using the vehicle and pedestrian bounding boxes, and this graph structure is input into the graph neural network to obtain the pedestrian-vehicle conflict probability.

[0032] Step S106: Based on the current road characteristics and the current environmental data corresponding to the road with mixed pedestrian and vehicle traffic, determine whether the dynamic speed limit conditions are met.

[0033] In one example, the dynamic speed limit conditions include multiple thresholds corresponding to the current road features and current environmental data. By comparing the magnitude of the current road features, current environmental data and the corresponding thresholds, it is determined whether dynamic speed limit optimization is triggered.

[0034] Step S108: If so, determine the initial dynamic speed limit data through the dynamic speed limit model.

[0035] The dynamic speed limit model includes a safe speed limit sub-model, a traffic saturation speed limit sub-model, and a pedestrian priority speed limit sub-model. The safe speed limit, traffic saturation speed limit, and pedestrian priority speed limit are determined through the above sub-models, respectively. Based on the above current environmental data, weather correction coefficients and visibility correction coefficients are determined. The minimum value among the safe speed limit, traffic saturation speed limit, and pedestrian priority speed limit is corrected using the weather correction coefficients and visibility correction coefficients to obtain the initial dynamic speed limit data.

[0036] Step S110: Determine the weather-corrected average vehicle speed based on vehicle status data and current environmental data, and use the weather-corrected average vehicle speed to correct the initial dynamic speed limit data to obtain the target dynamic speed limit data.

[0037] Among them, the weather-corrected average speed is the product of the average speed data in the vehicle status data and the weather correction coefficient and visibility correction coefficient; based on the weather-corrected average speed, pedestrian density, and probability of pedestrian-vehicle conflict, it is determined whether to relax, tighten, or maintain the above initial dynamic speed limit data to obtain the final target dynamic speed limit data.

[0038] The dynamic speed limit optimization method for mixed-traffic roads provided in this invention determines whether a speed limit adjustment is triggered based on current road features extracted from road visual data and current environmental data. If so, an initial dynamic speed limit is determined through a dynamic speed limit model. Then, the initial dynamic speed limit data is corrected using weather-corrected average vehicle speed to obtain the final target dynamic speed limit data. This invention can alleviate congestion on mixed-traffic roads, improve road communication capabilities, and ensure the traffic safety of pedestrians, non-motorized vehicles, and motorized vehicles by adjusting the speed limit in real time, thus achieving refined and intelligent management of speed limits on mixed-traffic roads.

[0039] This invention aims to address the challenges of balancing safety and efficiency, as well as lagging regulation, in roads where pedestrians and vehicles share the same space. By utilizing real-time data collection and intelligent speed limit optimization, it improves road traffic safety and efficiency. Dynamically adjusting speed limit strategies effectively reduces traffic accidents, alleviates traffic congestion, achieves precise management, and promotes green transportation and smart city development. Simultaneously, a feedback optimization mechanism is constructed to continuously improve system adaptability, providing an efficient and safe speed limit solution for intelligent transportation.

[0040] For ease of understanding, this invention provides a specific implementation method for optimizing dynamic speed limits on roads with mixed pedestrian and vehicle traffic. See [link to relevant documentation]. Figure 2 The flowchart shown is a different method for optimizing dynamic speed limits on roads with mixed pedestrian and vehicle traffic, including: Step 1, Install detection equipment: Install a radar-visual integrated machine every 100 meters on straight sections of road, and install cameras at intersections to cover lanes and pedestrian areas.

[0041] (1.1) Equipment Spacing and Quantity Design: On straight sections of road, one integrated radar-visual unit shall be installed every 100 meters to ensure monitoring coverage of each lane and adjacent areas. At intersections, at least two high-definition cameras shall be installed in each direction to ensure comprehensive monitoring of the intersection area. For complex intersections, the number of equipment should be increased by 50% based on the monitoring coverage of the intersection.

[0042] (1.2) Equipment types and technical specifications: The RayVision All-in-One Machine provides integrated capabilities for both LiDAR and camera data capture. It boasts a resolution of at least 2K and supports dynamic motion trajectory capture. The maximum recognition range of the RayVision All-in-One Machine covers a radius of 80 meters. Its real-time data refresh rate is at least 10Hz.

[0043] Step 2, dividing the monitoring area: using the Mask R-CNN algorithm of the camera to perform semantic segmentation of the image, dividing it into vehicle traffic area and pedestrian traffic area.

[0044] In one implementation, the vehicle and pedestrian zones of a road with mixed pedestrian and vehicle traffic are identified based on road visual data. Specifically: Vehicle traffic area: Lane boundaries are marked using image semantic segmentation algorithms, and ROI (Region of Interest, which refers to a specific area selected in image or video processing for centralized analysis and processing of data) is defined to facilitate the calculation of vehicle flow and speed.

[0045] Pedestrian Area: The edges of the pedestrian walkway are automatically identified from camera data using an image semantic segmentation algorithm (MaskR-CNN, used to identify the boundaries between vehicular and pedestrian areas).

[0046] Step 3, Road Feature Extraction: Extract one or more of the following based on the vehicle and pedestrian areas: vehicle status data, pedestrian status data, road occupancy rate, and probability of vehicle-pedestrian conflict.

[0047] (3.1) Perform target detection on the vehicle area and the pedestrian area respectively to obtain vehicle bounding boxes and pedestrian bounding boxes, and determine vehicle state data and pedestrian state data based on the vehicle bounding boxes and pedestrian bounding boxes.

[0048] In one implementation, vehicle status data includes traffic flow and vehicle speed data. Traffic flow ( The system uses the YOLOv8 algorithm to detect vehicle types and count the number of vehicles passing through, uploading the vehicle count to the server and calculating the traffic flow per unit time per second. Vehicle speed ( The system uses a radar module to measure vehicle speed in real time, and combines this with optical flow for speed verification. In practice, the data is corrected by a spatiotemporal synchronization module to eliminate static obstacle interference.

[0049] In one implementation, pedestrian state data includes at least pedestrian density ( Specifically, it includes: (I) The YOLOv8 target detection algorithm is used to detect pedestrians in real-time video streams or image sequences captured by the camera, and pedestrian bounding boxes and confidence scores are output. Redundant detection boxes are filtered by non-maximum suppression (NMS) and detection results with confidence scores higher than the threshold are retained. Based on the DeepSORT multi-object tracking algorithm, pedestrian appearance features and motion information are extracted, cross-frame continuous trajectories are generated and assigned unique identifiers (IDs). The position coordinates, IDs and motion state parameters of each pedestrian in each frame are output.

[0050] (II) Based on the camera calibration parameters, the pedestrian position in the image coordinate system is converted into the ground coordinate in the world coordinate system through the homography matrix; the pedestrian area is divided into grid units of a preset size (such as 1m×1m), the number of pedestrians in each grid is counted, and the initial pedestrian density in each grid is obtained.

[0051] (III) The DBSCAN clustering algorithm is used to cluster the grids according to the initial pedestrian density in each grid, so as to divide the crowded area, the non-crowded area and their corresponding target pedestrian density.

[0052] (3.2) Determine the road occupancy rate based on the boundary information of the vehicle area and the pedestrian area.

[0053] In one implementation, extracting relevant areas from the vehicular and pedestrian areas includes: Occupied Area: Extract the area data of vehicle / pedestrian bounding boxes and accumulate the occupied area of ​​all vehicles and pedestrians.

[0054] Total road area: The road model is reconstructed using point cloud data from LiDAR, and the area value of the coverage area is dynamically obtained.

[0055] Use the formula: ,in, For road occupancy, This refers to the total area occupied by all vehicles and pedestrians. This represents the total area of ​​the monitored region.

[0056] (3.3) The probability of human-vehicle conflict is determined based on vehicle bounding boxes and pedestrian bounding boxes using a graph neural network.

[0057] In this embodiment of the invention, a graph neural network (GNN) is used to automatically identify pedestrian-vehicle conflict events. This is used to determine whether a pedestrian-vehicle conflict has occurred in each frame of monitoring data on roads with mixed pedestrian and vehicle traffic, and to count the number of conflicts within a one-minute time window. This increases the probability of subsequent conflicts between pedestrians and vehicles. The calculation provides the basis. In one implementation, the probability of a pedestrian-vehicle conflict can be determined by following these steps. : (I) Data Labeling: Frame-level global binary classification labeling is used. Specific vehicles / pedestrians, bounding boxes, and relationships are not labeled; only "whether there are any conflicts in this frame" is indicated. The labeling rules are as follows: Label 0 indicates non-conflict, including: pedestrians and vehicles traveling in their own lanes with no dangerous interactions; pedestrian-vehicle distance ≥ 5 meters; no sudden braking, no cutting in, and no non-motorized vehicles encroaching on the lane.

[0058] Label 1 indicates a conflict. A conflict is indicated if any of the following conditions are met: a pedestrian / non-motorized vehicle enters the motor vehicle lane and is less than 5 meters away from the vehicle; the vehicle brakes suddenly to avoid the pedestrian; the pedestrian is close to the edge of the lane and has a clear tendency to encroach on the lane; a motor vehicle and a non-motorized vehicle pass each other at close range and there is a risk of collision.

[0059] (II) Use vehicle bounding boxes as vehicle nodes and pedestrian bounding boxes as pedestrian nodes (that is, define each vehicle and each pedestrian as an independent node), and assign node features to vehicle nodes and pedestrian nodes to obtain the initial mixed human and vehicle graph structure. The initial node features include one or more of the following: speed features, position features, movement direction features, and node type features (vehicle or pedestrian).

[0060] (III) When the distance between any two nodes is less than a preset threshold, establish an edge between the two nodes to obtain the intermediate pedestrian-vehicle mixed traffic graph structure. Here, the edge between each pair of nodes represents their interaction relationship. For example, when the distance between two vehicles / pedestrians / vehicles is less than 5 meters, it is defined as the existence of an interaction edge, thus obtaining the adjacency matrix. And define the intermediate human-vehicle mixed graph structure with interactive edges.

[0061] (IV) Perform at least one message pass on the intermediate mixed-traffic graph structure to determine the node aggregation features of vehicle nodes and pedestrian nodes, and obtain the target mixed-traffic graph structure.

[0062] In one implementation, the target pedestrian-vehicle mixed traffic graph structure is obtained by performing operations such as message passing, node feature updating, and global graph embedding on the intermediate pedestrian-vehicle mixed traffic graph structure. Specifically: Message Passing: Each node passes its own feature information to neighboring nodes through edges and receives feature information from neighboring nodes, realizing feature interaction between nodes; ; in, , For nodes In the , The feature vector of the layer, For activation function, The weight matrix is ​​a learnable matrix. For nodes The set of neighboring nodes, For nodes In the The feature vector of the layer, The edge weight is denoted as .

[0063] Node feature update: Through multiple message passes, the feature vector of each node will gradually converge with global information, accurately reflecting its interaction relationship with other nodes; Global graph embedding: Finally, by aggregating the feature information of all nodes and edges, the feature representation of the entire graph is obtained, and classification is performed to identify whether there are conflict events.

[0064] (V) Generate conflict event recognition results based on the target human-vehicle mixed traffic graph structure using graph neural networks.

[0065] In one implementation, the graph neural network structure is as follows: Input: Node features and adjacency matrix ; GCN Layer 1: Input: 4-dimensional features; Output: 64-dimensional node embeddings; Activation: ReLU; GCN Layer 2: Input: 64-dimensional node embedding; Output: 128-dimensional node embedding; Activation: ReLU; Global graph embedding (global average pooling): The 128-dimensional features of all nodes are averaged to obtain a 1×128-dimensional vector, which is the global feature of the entire graph, i.e., the global graph embedding above.

[0066] Classification Header (Fully Connected Layer): Input: 128-dimensional global features; Output: 1 probability value (0~1); Activation: sigmoid output, if the probability > 0.5 there is a conflict, if the probability ≤ 0.5 there is no conflict.

[0067] In this embodiment of the invention, GCN is a binary classification task, employing a binary cross-entropy loss function: the loss calculation formula is as follows: ; in Label 0 / 1. The model predicts the conflict probability, which is the output value of the sigmoid function.

[0068] (VI) Count the number of conflict events in the conflict event identification results, and use the ratio between the number of conflict events and the total number of conflict event identification results as the probability of human-vehicle conflict.

[0069] In one implementation, after the model is trained, it can be used to determine conflict events, set a time window, and calculate the probability of a conflict occurring within the window, including the probability of a human-vehicle conflict. Formula calculation: ; The number of pedestrian-vehicle conflict events detected within a time window (e.g., the window time is set to 1 minute, the GCN model identifies whether there is a conflict event every 5 seconds, and the sum of the number of conflicts within 1 minute is the value). ); This represents the total number of detections within the monitored area within the time window. (Probability of pedestrian-vehicle conflict) When the set threshold is exceeded, the system will prioritize triggering the speed limiting policy.

[0070] Step 4, determine the speed limit conditions: determine whether the speed limit is triggered based on the current road characteristics, and the conditions include at least the probability of conflict, pedestrian density and weather threshold.

[0071] In one implementation, the dynamic speed limit condition is determined to be met if the current road characteristics and the current environmental data corresponding to the mixed pedestrian and vehicle road meet any one of the following sub-conditions: Sub-condition 1: The probability of a pedestrian-vehicle conflict is greater than the first preset probability threshold. The first preset probability threshold The value typically ranges from 0.1 to 0.2 (i.e., a conflict probability of 10% to 20%).

[0072] Sub-condition 2: The speed difference between the maximum and minimum vehicle speed data in the vehicle status data is greater than a preset speed difference threshold. ;in, The speed difference This is the maximum vehicle speed data. This is the minimum vehicle speed data. The preset speed difference threshold is typically 10–20 km / h.

[0073] Sub-condition 3: The pedestrian density in the pedestrian status data is greater than the first preset density threshold. First preset density probability The typical value is 0.5–1.0 people / m².

[0074] Sub-condition 4: Snowfall or rainfall intensity data in the current environmental data Greater than the preset intensity threshold Among them, the preset intensity threshold It is set according to the drainage capacity and vehicle traction of the road design. The speed is typically 5–10 mm / h.

[0075] Sub-condition 5: Visibility data in the current environmental data less than the preset visibility threshold Among them, the preset visibility threshold It is determined based on standard traffic safety requirements, and is generally 200–500 meters.

[0076] Step 5: Generate a dynamic speed limit scheme: Taking into account safety, road saturation, and pedestrian priority principles, generate target dynamic speed limit data. The dynamic speed limit model includes a safe speed limit sub-model, a traffic saturation speed limit sub-model, and a pedestrian priority speed limit sub-model.

[0077] Specifically, it includes: (5.1) Determine the weather correction factor and visibility correction factor based on the current environmental data.

[0078] Weather correction factor The determination process is as follows: Gas correction factor The range of values ​​is 0 < ≤1, the calculation formula is: ; in: Rainfall intensity, in mm / h; This is the rainfall intensity influence coefficient, which can be set from 0.02 to 0.05, with a default value of 0.03; in snowy weather, it can be directly set to... Set it to 0.4.

[0079] Example: If the rainfall intensity is 10 mm / h, .

[0080] Visibility correction factor The determination process is as follows: Visibility correction factor The range of values ​​is 0 < ≤1, the calculation formula is: = ; in Standard visibility (1 km) under ideal conditions. This represents the actual visibility.

[0081] Example: If visibility drops to 500 meters, =0.5.

[0082] (5.2) Determine the safe speed limit, the flow saturation speed limit, and the pedestrian priority speed limit respectively through the safe speed limit sub-model, the flow saturation speed limit sub-model, and the pedestrian priority speed limit sub-model.

[0083] For the safety-limited speed sub-model: ; It is the coefficient of adhesion on wet and slippery surfaces. It is generally 0.3 to 0.5 in rainy weather and about 0.2 in snowy weather.

[0084] This is a sub-model for speed limits based on traffic saturation, calculated using road occupancy and weather factors, reflecting the required traffic efficiency under congested conditions. ; Free-flow speed refers to the maximum safe driving speed that a motor vehicle can maintain when there is no traffic / pedestrian interference and road conditions are good (no lane occupation, no construction, and normal weather). It can be referenced from the speed limit requirements for urban slow-moving mixed traffic areas in traffic engineering specifications. For example, the speed limit for motor vehicles in residential areas / school areas / commercial areas is generally 15~30km / h, and the lower value should be taken based on the scenario.

[0085] This is a pedestrian-priority speed limit sub-model, set based on pedestrian density, reflecting the core requirements for pedestrian safety; when When people / m², take This will further enhance the level of pedestrian safety.

[0086] The values ​​of all core parameters in the model follow traffic standards or design specifications for roads with mixed pedestrian and vehicle traffic. The specific value ranges and default values ​​are shown in Table 1 below: Table 1. Range of values ​​for all parameters in the model

[0087] (5.3) Using weather correction factors and visibility correction factors, the minimum value among the safe speed limit, traffic saturation speed limit, and pedestrian priority speed limit is corrected to obtain the initial dynamic speed limit data. Specifically, the initial dynamic speed limit data is determined according to the following formula: ) ; Constraints: Ensure that the speed limit is not lower than the minimum safe speed limit for roads where pedestrians and vehicles share the road.

[0088] Step 6, Publish speed limits and optimizations: Publish speed limit plans through road displays and traffic lights, monitor the implementation effect in real time, and dynamically adjust and optimize parameters.

[0089] In practical implementation, dynamically adjusting and optimizing parameters means: adjusting the average vehicle speed according to weather conditions. If the speed limit is too low, relax it appropriately. If pedestrian density or the probability of conflict increases significantly, further tighten the speed limit and issue warning signals. (Based on weather-corrected average vehicle speed after environmental adjustments.) pedestrian density Probability of pedestrian-vehicle conflict The real-time changes in the speed limit are quantified, and the adjustment time window is set to 1 minute to avoid frequent adjustments that could lead to driver misjudgment. The specific rules are as follows: (6.1) The product of the average vehicle speed data in the vehicle status data, the weather correction factor, and the visibility correction factor is taken as the weather-corrected average speed. Specifically, the weather-corrected average speed can be determined using the following formula. : ; in, This refers to vehicle speed data within the vehicle status data. This represents the total number of vehicle speed data points in the vehicle status data.

[0090] (6.2) Determine the current vehicle speed threshold based on the initial dynamic speed limit data, for example, by... As the current vehicle speed threshold.

[0091] (6.3) Relaxing speed limits: When the weather-corrected average vehicle speed is less than the current vehicle speed threshold, the probability of pedestrian-vehicle conflict is less than the second preset probability threshold, and the pedestrian density in the pedestrian status data is less than the second preset density threshold, the initial dynamic speed limit data is adjusted according to the preset increment until the weather-corrected average vehicle speed reaches the preset road free-flow speed.

[0092] For example, when monitored ,and people / m² Each time, the speed limit is increased by 5 km / h until the free-flow speed of the road is reached. And even after the speed limit is relaxed, it shall not exceed the fixed speed limit of the roadbed.

[0093] (6.4) Tighten speed limit: When the probability of pedestrian-vehicle conflict is greater than the third preset probability threshold and the pedestrian density in the pedestrian status data is greater than the third preset density threshold, the initial dynamic speed limit data is adjusted according to the preset reduction until the adjusted speed limit data reaches the preset minimum safe speed limit.

[0094] For example, when monitored people / m² or Each time, the speed will be reduced by 5 km / h until the minimum safe speed limit is reached. At the same time, road warning signals are issued.

[0095] (6.5) Maintain speed limit: Except for the two cases mentioned above, the system maintains the current dynamic speed limit value. constant.

[0096] Furthermore, this embodiment of the invention also provides dynamic optimization of weather / visibility coefficients: based on real-time rainfall, snow cover, and visibility data, the calculation parameters of the environmental coefficients are dynamically adjusted to make the coefficients more closely match the actual environmental conditions. The specific optimization rules are as follows: When the rainfall intensity exceeds 15 mm / h for 10 consecutive minutes, the rainfall intensity influence coefficient will be... The weather coefficient was increased from 0.03 to 0.05, thus lowering the weather coefficient. When the rainfall intensity drops below 5 mm / h, recovery... To the initial value of 0.03; When visibility is less than 200 meters for 5 consecutive minutes, The original calculation will be reduced by another 20%; when visibility recovers to over 500 meters, the recovery... Return to the original calculated value; Safety incident feedback optimization: When the incident rate within the monitored area increases by 10% compared to the average of the past 7 days, The coefficient will be reduced by 20% from its original value; once the accident rate returns to its average value, the coefficient will be restored.

[0097] In summary, the primary technical problem addressed by this invention is how to generate accurate speed limit strategies based on multi-source data in real-time dynamic mixed-traffic scenarios, while simultaneously considering traffic efficiency and safety. This invention proposes a dynamic speed limit optimization method for mixed-traffic roads, specifically addressing the traffic management needs of such roads. This method is a crucial component of the signal control optimization module in a multi-objective intersection optimization system, aimed at improving vehicle throughput. It seeks to solve the problems of low throughput, significant safety hazards, and the outdated nature of traditional speed limit methods in mixed-traffic scenarios. 1) Improve traffic efficiency: By dynamically adjusting vehicle speed limits in real time, alleviate congestion on roads with mixed pedestrian and vehicle traffic and improve road capacity; 2) Reduce traffic risks: Optimize speed limit strategies to ensure traffic safety for pedestrians, non-motorized vehicles, and motorized vehicles; 3) Optimize traffic efficiency: Alleviate traffic congestion and improve road capacity; 4) Adapt to multiple scenarios: Applicable to complex weather and various road types; 5) Energy conservation and emission reduction: Reduce fuel consumption and pollution, and protect the environment; 6) Achieve intelligent management: Based on multi-source data collection and real-time analysis, achieve refined and intelligent management of speed limits on mixed-traffic roads.

[0098] The implementation of this invention will significantly improve the traffic operation of roads where pedestrians and vehicles share the road, promote the intelligent and efficient development of urban transportation, and provide technical support for the field of intelligent transportation.

[0099] Based on the foregoing embodiments, this invention provides a dynamic speed limit optimization device for roads with mixed pedestrian and vehicle traffic. Multiple visual sensors are deployed in designated areas of these roads. (See also...) Figure 3 The diagram shows a structural schematic of a dynamic speed limit optimization device for roads with mixed pedestrian and vehicle traffic. The device mainly includes the following parts: Data acquisition module 302 is used to acquire road visual data collected by the visual sensor for roads with mixed pedestrian and vehicle traffic; The feature extraction module 304 is used to extract current road features from road visual data. The current road features include one or more of vehicle status data, pedestrian status data, road occupancy rate, and vehicle-pedestrian conflict probability. The vehicle-pedestrian conflict probability is obtained by processing the road visual data through a graph neural network. The judgment module 306 is used to determine whether the dynamic speed limit conditions are met based on the current road characteristics and the current environmental data corresponding to the road where pedestrians and vehicles share the road. Speed ​​limit determination module 308 is used to determine the initial dynamic speed limit data through the dynamic speed limit model when the result of the judgment module is yes. The speed limit correction module 310 is used to determine the weather-corrected average vehicle speed based on vehicle status data and current environmental data, and to correct the initial dynamic speed limit data using the weather-corrected average vehicle speed to obtain the target dynamic speed limit data.

[0100] The dynamic speed limit optimization method for mixed-traffic roads provided in this invention determines whether a speed limit adjustment is triggered based on current road features extracted from road visual data and current environmental data. If so, an initial dynamic speed limit is determined through a dynamic speed limit model. Then, the initial dynamic speed limit data is corrected using weather-corrected average vehicle speed to obtain the final target dynamic speed limit data. This invention can alleviate congestion on mixed-traffic roads, improve road communication capabilities, and ensure the traffic safety of pedestrians, non-motorized vehicles, and motorized vehicles by adjusting the speed limit in real time, thus achieving refined and intelligent management of speed limits on mixed-traffic roads.

[0101] In one implementation, the feature extraction module 304 is specifically used for: Based on road visual data, identify the vehicle and pedestrian zones on roads with mixed pedestrian and vehicle traffic, and extract one or more of the following current road features: Target detection is performed on the vehicle area and the pedestrian area respectively to obtain vehicle bounding boxes and pedestrian bounding boxes, and vehicle state data and pedestrian state data are determined based on the vehicle bounding boxes and pedestrian bounding boxes; Road occupancy is determined based on the boundary information of vehicle traffic areas and pedestrian areas; The probability of human-vehicle conflict is determined based on vehicle bounding boxes and pedestrian bounding boxes using a graph neural network.

[0102] In one implementation, the feature extraction module 304 is specifically used for: Using vehicle bounding boxes as vehicle nodes and pedestrian bounding boxes as pedestrian nodes, and assigning node features to vehicle nodes and pedestrian nodes, an initial mixed-traffic graph structure is obtained. The initial node features include one or more of the following: speed features, position features, movement direction features, and node type features. If the distance between any two nodes is less than a preset threshold, an edge is established between the two nodes to obtain the intermediate pedestrian and vehicle mixed traffic graph structure. Perform at least one message pass on the intermediate human-vehicle mixed traffic graph structure to determine the node aggregation characteristics of vehicle nodes and pedestrian nodes, and obtain the target human-vehicle mixed traffic graph structure. Conflict event identification results are generated based on the target pedestrian-vehicle mixed traffic graph structure using a graph neural network. The number of conflict events in the conflict event identification results is counted, and the ratio between the number of conflict events and the total number of conflict event identification results is used as the probability of human-vehicle conflict.

[0103] In one implementation, the determination module 306 is specifically used for: If the current road characteristics and the current environmental data corresponding to the mixed pedestrian and vehicle road meet any one of the following sub-conditions, the dynamic speed limit condition is determined to be met: Sub-condition 1: The probability of a human-vehicle conflict is greater than the first preset probability threshold; Sub-condition 2: The speed difference between the maximum and minimum vehicle speed data in the vehicle status data is greater than a preset speed difference threshold; Sub-condition 3: The pedestrian density in the pedestrian status data is greater than the first preset density threshold; Sub-condition 4: The snowfall or rainfall intensity data in the current environmental data is greater than the preset intensity threshold; Sub-condition 5: The visibility data in the current environment is less than the preset visibility threshold.

[0104] In one implementation, the dynamic speed limit model includes a safe speed limit sub-model, a traffic saturation speed limit sub-model, and a pedestrian priority speed limit sub-model; the speed limit determination module 308 is specifically used for: Based on the current environmental data, determine the weather correction factor and the visibility correction factor respectively; The safe speed limit, the flow saturation speed limit, and the pedestrian priority speed limit are determined by the safe speed limit sub-model, the flow saturation speed limit sub-model, and the pedestrian priority speed limit sub-model, respectively. By using weather correction factors and visibility correction factors, the minimum values ​​among the safe speed limit, traffic saturation speed limit, and pedestrian priority speed limit are corrected to obtain the initial dynamic speed limit data.

[0105] In one implementation, the speed limit correction module 310 is specifically used for: The weather-corrected average speed is calculated by multiplying the average speed data in the vehicle status data, the weather correction factor, and the visibility correction factor.

[0106] In one implementation, the speed limit correction module 310 is specifically used for: Determine the current vehicle speed threshold based on initial dynamic speed limit data; If the weather-corrected average vehicle speed is less than the current vehicle speed threshold, the probability of pedestrian-vehicle conflict is less than the second preset probability threshold, and the pedestrian density in the pedestrian status data is less than the second preset density threshold, the initial dynamic speed limit data will be adjusted according to the preset increment until the weather-corrected average vehicle speed reaches the preset road free-flow speed. If the probability of a pedestrian-vehicle conflict is greater than the third preset probability threshold, and the pedestrian density in the pedestrian status data is greater than the third preset density threshold, the initial dynamic speed limit data will be adjusted according to the preset reduction until the adjusted speed limit data reaches the preset minimum safe speed limit.

[0107] The device provided in this embodiment of the invention has the same implementation principle and technical effect as the aforementioned method embodiment. For the sake of brevity, any parts not mentioned in the device embodiment can be referred to the corresponding content in the aforementioned method embodiment.

[0108] This invention provides an electronic device, specifically, the electronic device includes a processor and a memory; the memory stores a computer program, which, when run by the processor, executes the method described in any of the above embodiments.

[0109] Figure 4 The present invention provides a schematic diagram of the structure of an electronic device 100, which includes a processor 40, a memory 41, a bus 42 and a communication interface 43. The processor 40, the communication interface 43 and the memory 41 are connected through the bus 42. The processor 40 is used to execute executable modules, such as computer programs, stored in the memory 41.

[0110] The memory 41 may include high-speed random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 43 (which can be wired or wireless), such as the Internet, wide area network, local area network, metropolitan area network, etc.

[0111] Bus 42 can be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 4 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0112] The memory 41 is used to store programs. After receiving an execution instruction, the processor 40 executes the program. The method executed by the device for defining the flow process disclosed in any of the foregoing embodiments of the present invention can be applied to the processor 40 or implemented by the processor 40.

[0113] Processor 40 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of processor 40 or by instructions in software form. Processor 40 can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this invention. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly manifested as execution by a hardware decoding processor, or execution by a combination of hardware and software modules in the decoding processor. The software module can reside in a mature storage medium in the field, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory 41, and the processor 40 reads the information from memory 41 and, in conjunction with its hardware, completes the steps of the above method.

[0114] The computer program product of the readable storage medium provided in the embodiments of the present invention includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods described in the foregoing method embodiments. For specific implementation, please refer to the foregoing method embodiments, which will not be repeated here.

[0115] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0116] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for optimizing dynamic speed limits on roads with mixed pedestrian and vehicle traffic, characterized in that, Multiple visual sensors are deployed in designated areas of roads where pedestrians and vehicles share the road. The method includes: Acquire the road visual data collected by the visual sensor for the road where pedestrians and vehicles share the road; The current road features are extracted from the road visual data. The current road features include one or more of vehicle status data, pedestrian status data, road occupancy rate, and probability of pedestrian-vehicle conflict. The probability of pedestrian-vehicle conflict is obtained by processing the road visual data through a graph neural network. Based on the current road characteristics and the current environmental data corresponding to the mixed pedestrian and vehicle road, determine whether the dynamic speed limit conditions are met. If so, the initial dynamic speed limit data is determined through the dynamic speed limit model; Based on the vehicle status data and the current environment data, the weather-corrected average speed is determined, and the initial dynamic speed limit data is corrected using the weather-corrected average speed to obtain the target dynamic speed limit data.

2. The method for optimizing dynamic speed limits on roads with mixed pedestrian and vehicle traffic according to claim 1, characterized in that, Extracting current road features from the road visual data includes: Based on the road visual data, the vehicle and pedestrian areas of the mixed-traffic road are identified to extract one or more of the following current road features: Target detection is performed on the vehicle area and the pedestrian area respectively to obtain vehicle bounding boxes and pedestrian bounding boxes, and vehicle state data and pedestrian state data are determined based on the vehicle bounding boxes and pedestrian bounding boxes. The road occupancy rate is determined based on the boundary information of the vehicle traffic area and the pedestrian area. The probability of a human-vehicle conflict is determined based on the vehicle bounding box and the pedestrian bounding box using a graph neural network.

3. The method for optimizing dynamic speed limits on roads with mixed pedestrian and vehicle traffic according to claim 2, characterized in that, Using a graph neural network, the probability of pedestrian-vehicle conflict is determined based on the vehicle bounding box and the pedestrian bounding box, including: Using the vehicle bounding box as vehicle nodes and the pedestrian bounding box as pedestrian nodes, and assigning node features to the vehicle nodes and the pedestrian nodes, an initial mixed pedestrian and vehicle traffic graph structure is obtained. The initial node features include one or more of the following: speed features, position features, movement direction features, and node type features. If the distance between any two nodes is less than a preset threshold, an edge is established between the two nodes to obtain the intermediate pedestrian and vehicle mixed traffic graph structure. At least one message pass is performed on the intermediate mixed-traffic graph structure to determine the node aggregation features of the vehicle nodes and the pedestrian nodes, thereby obtaining the target mixed-traffic graph structure. The conflict event identification result is generated based on the target mixed pedestrian and vehicle traffic graph structure using a graph neural network. The number of times a conflict event occurs in the conflict event identification results is counted, and the ratio between the number of conflict events occurring and the total number of conflict event identification results is taken as the probability of human-vehicle conflict.

4. The method for optimizing dynamic speed limits on roads with mixed pedestrian and vehicle traffic according to claim 1, characterized in that, Based on the current road features and the current environmental data corresponding to the mixed pedestrian and vehicle road, determine whether the dynamic speed limit conditions are met, including: If the current road features and the current environmental data corresponding to the mixed pedestrian and vehicle road meet any one of the following sub-conditions, the dynamic speed limit condition is determined to be met: Sub-condition 1: The probability of a human-vehicle conflict is greater than a first preset probability threshold; Sub-condition 2: The speed difference between the maximum and minimum vehicle speed data in the vehicle status data is greater than a preset speed difference threshold; Sub-condition 3: The pedestrian density in the pedestrian status data is greater than the first preset density threshold; Sub-condition 4: The snowfall or rainfall intensity data in the current environmental data is greater than a preset intensity threshold; Sub-condition 5: The visibility data in the current environment data is less than the preset visibility threshold.

5. The method for optimizing dynamic speed limits on roads with mixed pedestrian and vehicle traffic according to claim 1, characterized in that, The dynamic speed limit model includes a safe speed limit sub-model, a traffic saturation speed limit sub-model, and a pedestrian priority speed limit sub-model. The initial dynamic speed limit data is determined through a dynamic speed limit model, including: Based on the current environmental data, determine the weather correction factor and the visibility correction factor respectively; The safe speed limit, the flow saturation speed limit, and the pedestrian priority speed limit are determined respectively through the safe speed limit sub-model, the flow saturation speed limit sub-model, and the pedestrian priority speed limit sub-model. Using the weather correction coefficient and the visibility correction coefficient, the minimum value among the safe speed limit, the traffic saturation speed limit, and the pedestrian priority speed limit is corrected to obtain the initial dynamic speed limit data.

6. The method for optimizing dynamic speed limits on roads with mixed pedestrian and vehicle traffic according to claim 5, characterized in that, Determining the weather-corrected average vehicle speed based on the vehicle status data and the current environmental data includes: The average speed is calculated by multiplying the average speed data in the vehicle status data, the weather correction factor, and the visibility correction factor.

7. The method for optimizing dynamic speed limits on roads with mixed pedestrian and vehicle traffic according to claim 1, characterized in that, The initial dynamic speed limit data is corrected using the weather-corrected average vehicle speed to obtain the target dynamic speed limit data, including: The current vehicle speed threshold is determined based on the initial dynamic speed limit data; If the weather-corrected average vehicle speed is less than the current vehicle speed threshold, the probability of pedestrian-vehicle conflict is less than the second preset probability threshold, and the pedestrian density in the pedestrian status data is less than the second preset density threshold, the initial dynamic speed limit data is adjusted according to a preset increment until the weather-corrected average vehicle speed reaches the preset road free-flow speed. If the probability of a pedestrian-vehicle conflict is greater than a third preset probability threshold and the pedestrian density in the pedestrian status data is greater than a third preset density threshold, the initial dynamic speed limit data is adjusted according to a preset reduction until the adjusted speed limit data reaches the preset minimum safe speed limit.

8. A dynamic speed limit optimization device for roads with mixed pedestrian and vehicle traffic, characterized in that, Multiple visual sensors are deployed in designated areas of roads where pedestrians and vehicles share the road. The device includes: The data acquisition module is used to acquire road visual data collected by the visual sensor for the road where pedestrians and vehicles share the road. The feature extraction module is used to extract current road features from the road visual data. The current road features include one or more of vehicle status data, pedestrian status data, road occupancy rate, and vehicle-pedestrian conflict probability. The vehicle-pedestrian conflict probability is obtained by processing the road visual data through a graph neural network. The judgment module is used to determine whether the dynamic speed limit conditions are met based on the current road features and the current environmental data corresponding to the mixed pedestrian and vehicle road. The speed limit determination module is used to determine the initial dynamic speed limit data through the dynamic speed limit model when the result of the judgment module is yes. The speed limit correction module is used to determine the weather-corrected average vehicle speed based on the vehicle status data and the current environment data, and to correct the initial dynamic speed limit data using the weather-corrected average vehicle speed to obtain the target dynamic speed limit data.

9. An electronic device, characterized in that, The method includes a processor and a memory, the memory storing computer-executable instructions executable by the processor, the processor executing the computer-executable instructions to implement the method of any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions that, when invoked and executed by a processor, cause the processor to perform the method according to any one of claims 1 to 7.

Images